Nuclear Fusion's New Idea: An Off-the-Shelf Stellarator
spectrum.ieee.org
spectrum.ieee.org
But even if you have a working stellarator that's a very long way from an economically viable energy source. You've still got to a) figure out how to cheaply convert the released energy into electricity (and the baseline way of doing that in D-T fusion is...a steam turbine), and b) figure out materials that can survive the radiation bombardment for a sufficiently long time.
In sunny places (and I fully acknowledge that's not all of the world) it's still going to be hard to beat sticking bits of special glass out in a field and connecting wires to it.
But we should sure as heck keep tinkering away at it!
In HN terms they are demonstrating a significantly faster REPL by keeping the project small and minimising use of esoteric or highly bespoke components.
It's the closest you can get to building your own stellarator by walking into radioshack. I think it's a pretty cool idea.
But in sunnier, warmer parts of the world (which notably includes India, Pakistan, Bangladesh, Indonesia, Nigeria, Egypt, Ethiopia, Iran, Mexico, and Brazil, amongst others), over the next few decades it's hard to see anything much competing against solar and batteries for the bulk of energy usage.
Though solar panels prefer if it isn't too hot. Still, any such inefficiencies are easily outweighed by their low cost. In Netherlands it is common to install practice (and advised) more Wp than the inverter can handle. Maybe also a solution if it gets too hot, make it up by having more panels.
Heat pumps work great under the right circumstances. In cold climates like Boston and north of it, ideally you want ground source, which means digging deep (or wide) holes to flow water through instead of just pulling heat from outside air. Alternatively, (or additionally, ideally) they're best used in homes with very tight air envelopes, good insulation, and heat/energy recovery ventilation (HRV/ERV) systems. Installed in radiant heating systems in well-built houses, heat pumps are fantastic and you'll be comfortable all year for pennies on the dollar.
The reality is that most (US) construction, especially older, is just terrible in terms of air seal and insulation. Couple that with a potentially undersized air-source heat pump, which gets inefficient as the outside temp gets near the low end of its operating range, and you will not have a great time. But even then they are a good way to supplement gas heating, so you can limit furnace use to the coldest weeks of the year and do the environment a favor, as well as cool in the summer.
Are you one of the grifters or just confused?
ACs in cold climates are cheap. Cooling a house down 20 degrees doesnt take a lot of BTUs. Plus they’re just cheaper per-BTU and a typical northeast homeowner is going to be comfortable installing a cheap brand because it’s not life-or-death if your AC goes out.
When you start talking about having a heat pump being the primary heat source in a cold climate home, that’s a different ball game. First you need WAY more BTUs to heat a house in winter lows — even NYC hits single-digits most winters, that’s a 60 degree temperature differential. And you need to be prepared for the worst temperatures which basically necessitates a top-end brand like Mitsubishi which is rated down to historic-ish lows.
Not to mention that many homes don’t have central ducts, so converting to a heat pump means ductwork or mini splits..
All in all it’s not really a negligible thing.
Here’s a concrete example that took 5 minutes to find. Two condensers from the same company, same seller, same SEER, same series, same tonnage, except the second does heat. The heat pump is 29% more expensive.
2 Ton 14.3 SEER2 Trane Air Conditioner Condenser - RT Series. $1615
2 Ton 14.3 SEER2 Trane Heat Pump Condenser - RT Series. $2090
https://hvacdirect.com/trane-a4ac4024d1-2-ton-14-3-seer2-air...
https://hvacdirect.com/trane-a4hp4024d1-2-ton-14-3-seer2-hea...
You can, right now, buy air source heat pumps rated to -28C (-14F). While not enough for record lows in, say, Chicago or Toronto, it's plenty for places like New York, Seattle, and Vancouver.
If anything having a backup heat source makes me feel secure. A lot of things can go wrong with a gas furnace; some things can go wrong with a heat pump; but 10kW of electric restive heat is dead simple and just by itself it can provide 34000 BTU/h, which is 60% of the output of my old gas furnace. I don't think the duty cycle of my gas furnace ever exceeded 60% running time.
The auxiliary heater consists of two 5kW heaters, so even if one fails, there is still the other.
Do you have a wood burning back up? At my house in Massachusetts, we had a wood-burning stove that we could always fire up in a pinch. It never came to that, but it felt good knowing that we can always just burn a super hot fire and not freeze to death.
I live in an urban environment, so I'm not too worried about losing power for an extended period of time, and if I do I would probably just leave temporarily.
In my rural living fantasy I have a Ford F150 Lightning which I use as a "portable" backup battery that can be recharged by driving it to a Level 3 charging station. That and either a pellet stove and/or a Masonry Heater fireplace that sits opposite to a large equator (south) facing window.
I don't know much about how heat pump performance degrades, but going in to a life-and-death situation where the main heat system is operating outside rated performance seems like poor planning.
The newer ones can. Though that's partly my dislike. There's always yet another new technological solution for the various drawbacks. Which often are unproven (e.g. SMRs) and that often results in crazy cost overruns.
Nuclear has a crazy high fixed cost so ideally you still run it continuously.
Now whether it is optimal economically is another question. If you have some sources of energy in your grid that cost per usage (eg fossil fuels), you should rather switch those off than nuclear which costs the same whether you use it or not. But if your grid is almost all nuclear (eg France), you do load following.
Why would that need to be? The full needs need to be taken into account. But that's a TCO calculation, not something to add to the solar cost.
Nuclear energy in Europe tends to be way more expensive than initially budgeted. Resulting in a crazy difference in the kWh cost vs solar/wind. And there's more ways to store "electricity" than just batteries.
I do not see that implication.
Solar and wind is significantly cheaper than nuclear. That doesn't mean it is a replacement. That implies that there's a great way to solve the obvious drawbacks solar and wind have because of the high cost difference, plus speed/time things can be built.
A country needs to figure out their TCO and the energy mix. Which means yeah, the volatility needs to be solved. Which means that there needs to be more than solar/wind. But at the same time, nobody wants to invest in nuclear. It isn't commercially viable. It is important to not have to high electricity prices. Wikipedia has quite a section on the newest nuclear power station in the UK and the kWh cost for consumers at https://en.wikipedia.org/wiki/Hinkley_Point_C_nuclear_power_.... The cost was initially estimated at £24 per MWh but could now be £92.5.
Nuclear power stations have crazy cost overruns. The initial estimations are far too low (except maybe in China because they have recent experience).
> And if you need to do that then you need either a bunch of alternative on demand sources of energy (UK is using LNG) or some big storage capacity (+overprovision to fill them when it is sunny/windy). Nuclear doesn't need that.
Nuclear doesn't need a backup? UK built nuclear and has LNG. It's not so black and white as you stated.
>It’s economically viable to place them further north and in cloudier climates now
Which it really isn't unless we're neglecting to count in the need for long-term backup power. There still exists some amount of solar even now, though government subsidies and government provided peaker plants may have something to do with it
But that's not what the topic, no? It was about economically viable. It wasn't about fully relying on them. If someone can buy or place a solar panels/plant and it is economically feasible, then it is.
E.g. EU is finally connecting the various electrical grids together. So that electricity can more easily be exported/imported. Yet another way to deal with the fluctuations.
"then it is" what? economically viable? He still needs 100% backup from other sources, so you have to factor that cost in. Possible? Yeah, it's possible, but that wasn't the question.
Nope, you do not need to factor that in. If I put solar panels on my house, I check if it's worth the cost/investment. I am not intending to go fully off-grid, that is not the aim. Similarly for a solar plant. If it's economically viable it means if there's a good return on investment.
> He still needs 100% backup from other sources, so you have to factor that cost in.
Again, if you put solar panels up or if you have a solar plant it does not make any sense to factor in such costs to make a business case/economical sense. You're adding complications that aren't there.
I'm just not ignoring externalities. You can't do your thing _unless_ the other thing is being taken care of, you have a hard dependency on it, so factoring in the cost of that is the right thing to do. You'll pay that cost, too, be it via your net-hookup fees, or taxes that subsidize it. If you're lucky, others will pay more than you do and you can make more money. That's economically viable at a small scale but does not scale far, because you quickly run out of other people who foot the bill.
Much like tax havens do not scale, because they don't produce anything of value, their concept does not work without other countries where the value is being created.
European grids have been connected for decades, way before new renewables were a thing.
Agreed, but not in the amount of capacity that they're aiming for now due to increased (expected) volatility. See e.g. https://energy.ec.europa.eu/topics/infrastructure/electricit... which sets a 2030 target of 15%.
It is not a perfect way to deal with fluctuations, and it was proven beyond doubt that voters will not accept power prices to run unchecked in a EU connected grid.
Also, the various times I noticed talk about solar panels in the US it seems it is way overpriced, coupled with more expensive options being chosen. E.g. loads of micro inverters where a string inverter would make way more sense (economical).
Not affiliated, but e.g. https://www.solar-bouwmarkt.nl/ (use Google Translate), how do those prices compare to what you're seeing locally for panels and string inverters?
Only in the US, Australia, and few other places it makes sense to just put up some panels for free energy. Incidentally also sometimes apply to EV arguments.
The other half needs energy too though! And high-latitude areas are known to have dense enough populations and exceedingly high economic productivity, from Sweden to Ireland to New England.
Use the extra electricity to power machines that hydrogenate CO2 extracted from the atmosphere and turn it into methane.
Methane generated this way, (being the principal content of natural gas), would allow us to deliver nearly carbon-free fossil fuels to you and people in your biome, and everyone wins.
https://ethz.ch/en/news-and-events/eth-news/news/2024/08/iro...
I think OpenAI is investing into a fusion design that avoids steam for exactly this reason, so it's not just an anti-nuclear talking point.
Why not both? Can't we utilize the special glass to fetch energy from a man-made fusion reactor, in similar ways as we use it to fetch energy from the natural fusion reactor in space?
D + T or D + D fusion reaction produces neutrons and hard gamma radiation. The special glass cannot capture the energy of these directly.
The natural fusion reactor has a blanket that is literally thousands of kilometers thick. It effectively converts much of this into longer-wavelength electromagnetic radiation, from ultraviolet to infrared, with quite some visible light. That's what the special glass can make use of.
Also, the highly radioactive blanket is kept hundreds of millions km away from consumers, which helps alleviate the problem of disposing of radioactive waste. With a planet-based fusion reactor, we'd have to think where to put thousands and thousands of tons of slightly radioactive concrete which would realistically serve as a blanket.
I do think economic viability will be a major problem though. The fusion hype crowd focus on Q>1 being their major milestone but that's still a long way away from it being profitable to operate a fusion plant.
Of course, ammonia is chemically active, and water at 700K is also chemically active, so the turbines, as usual, require special metallurgy and regular maintenance.
The capital cost of just the turbines is enough to make it hard to compete with solar and batteries in many situations.
How many more trillion $$$ are going to be pissed away on this?
I'd say, shelve the idea for 50 to 75 years and then look at it again.
In the mean time, I think we could make major headway on a global elecrtic grid, that connects whatever part of the planet is sunny with all the rest. Add to this some major storage capacity, and I think we could resolve almost all of our energy problems with the money that would be wasted on further fusion efforts.
> In terms of its ability to confine particles, Muse is two orders of magnitude better than any stellarator previously built
Is it? It doesn't seem as if they have reached first plasma or have plans to do so anytime soon. Using electromagnets to not only confine but also to control the the plasma is a big selling point of the stellarator design, and they don't seem to address this.
This seems really cool, and I love the idea of lower-cost fusion. (Or even just functional fusion.) There are about a dozen companies making real progress in fusion, but I can't quite figure out what this team has actually accomplished.
What am I missing?
This quote reminded of the SpaceX’s approach to engineering and why they have leapfrogged past Boeing. Instead of spending 10-20 years and billions into a single design, SpaceX iterates.
This is a plasma test stand, and because it is so simple, you can potentially quickly iterate through different field configurations. This is at least a little bit useful, because a full Stellerator is extremely complicated to take apart, so you can't just change the coils around if you want to change something.
Roughly speaking energy can be mechanical, for particles or radiative, for photons. The first one is proportional to the temperature (the famous NRT) and the second is proportional to the fourth power of the temperature. The constant of proportionality is very small, and at regular temperatures we generally don't think of it that much. But at millions of degrees Kelvin, it starts to dominate all considerations.
The heat always moves form hot to cold. In the case of particles the heat flow is proportional to the difference in temperature, and in the case of radiation with the difference in temperature to the power 4. But heat also travels from particles to photons and vice-versa. It doesn't matter how.
The problem with fusion is now this. Suppose that you have a super-duper device, let's call it brompillator. It brings an amount of deuterium-tritium mix at the required temperature, let's say 10 million Kelvin. Now that volume of plasma is surrounded by cold stuff. You can imagine that you have some mirrors, or magnetic fields, or some magic stuff, but the cold hard stuff is that that plasma will want to radiate to the exterior and the flow of heat would be proportional to the surface area times the fourth power of the difference in temperature. Since for all practical purposes the outer temperature is zero, we are talking about the fourth power of 10 million Kelvin. Now that constant of porportionality is very small, it is called the Stefan-Boltzman constant and has a value of about 10^7 W m^-2 K^-4. Let's say the surface area is 1 square meter. So the heat loss happens at a rate of 10^-7 times (10^7)^4 = 10^21 Watts. That is 10^12 GigaWatts. One GW is the output of a decent sized nuclear power plant.
Of course, you can try to shield that plasma, but that shield has to be 99.99999....9% effective, where the number of 9s needs to be about 15 or so.
That is the immensity of the challenge that nobody is willing to tell you about.
How was this overcome in the case of the thermonuclear bomb? People imagine that once you have a fission bomb, you just put some deuterium-tritium mix next to it, and voila, you have a fusion bomb. No. The world's greatest minds have worked at this issue for about 5 years. The solution was something like this: if you first compress significantly the volume of fusion fuel, then the heat losses are much smaller (remember they are proportional to the area, and that's proportional to the square of the radius). They will still be tremendous, but you don't even aim to keep the reaction going for a long time. The duration of the fusion reaction in a thermonuclear bomb is still classified information, but public sources put it at the order of 1 microsecond. The heat losses are still tremendous, but for a short moment the heat gains from the fusion reaction are even greater, so ignition is achieved.
In the NIF experiment that achieved more than breakeven 2 years ago, the fusion lasted less than 10 nanoseconds [1].
If someone thinks the brompillator will achieve fusion and that will run for years, or even hours, or seconds, they don't understand the fundamental problem. Unfortunately, nobody is willing to ask hard questions about this, not even Sabine Hossenfelder.
[1] https://journals.aps.org/prl/pdf/10.1103/PhysRevLett.132.065...
I'm not sure that's even true, because if you manage to crack that, you still have the problem that your sustainable reaction is pumping out most of its energy in the form of very fast neutrons, which are (a) very hard to harvest energy from and (b) extremely bad for people and materials if you don't. You could have a self-sustaining reaction that you can't actually use!
Engineering studies of stellarators found they tend to be larger and have worse economics than tokamaks.
High energy neutrons leave the system. They cause damage to the container (ie neutron embrittlement) and that's a separate problem. But the real problem is the energy loss from the system.
Charged particles can be contained. Personally I think there are limits to even that because a high-temperature plasma is turbulent [1]. Containing that is just a hugely difficult problem.
I'm not convinced that nuclear fusion will ever be commercially viable.
All while we already have emission-free, reliable and cheap energy production in the form of solar power. [1]: https://www.psfc.mit.edu/research/topics/plasma-turbulence
Is there any chance you'd recommend any books related to these topics? The walk through decades of revelations in physics was the most enjoyable aspect of that book, I'd love to continue building on that story.
Here are some more books I read on this topic. One was written by someone who was very close to the Ulam and Teller inner circle: "Building the H Bomb" by Kenn Ford. Another one is "Sun in a Bottle: The Strange History of Fusion and the Science of Wishful Thinking" by Charles Seife. And finally, you can't go wrong with any book written by James Mahaffey.
Have I understood the consequences of those reports wrong? Does the heat loss you talk about only occur with fusion? (And if so, is it even a problem if the conditions for fusion to occur can be created by external heating this "easily"?)
But a thin metal sheet has no trouble doing this, as demonstrated by the Apollo lunar lander.
Some things are just not as hard as they sound. Magnetic confinement works very well. It easily achieves the necessary 9’s.
It’s just hard to keep it stable at millions of degrees, but that’s a different problem.
These things can fit so many nines of reliability.
So, when reports state that the a certain temperature was achieved and sustained for a certain period of time, what are they actually saying? We could go and find an article and get into some details, but I imagine they say that somewhere in the plasma that temperature was reached and sustained. But it is quite likely that that region is quite microscopic, maybe a very, very thin inner torus inside a larger torus. There is a gradient of temperature from the region where the announced temperature happens to the walls of the device. But one way or another that thin inner region can't have a surface area of anything close to 1 square meter. To get to 1 GW of power, you need 10^-12 square meters, and to get to 10 MW you need 10^-14 m2. That's about the surface area of a torus of (circular) length 3 m and diameter 1 femtometer. 1 femtometer is roughly the size of a nucleus of deuterium or tritium, so in principle this is the minimum diameter of a torus where you can talk about fusion.
So I did some searching, and found this stack exchange asking this question: https://physics.stackexchange.com/questions/415028/how-do-fu... . They argued that because fusion reactor plasma is optically thin, it doesn't radiate following blackbody radiation law. This textbook also say that: https://www.cambridge.org/core/books/abs/physics-of-plasmas/...
I tried to search more about plasma energy losses, and it becomes extremely complicated with insane amount of equations. One thing that I can get is that you can't model fusion reactor plasmas as a blackbody radiator because plasma is that complicated. If plasma is simpler then we should either have fusion already or we have given up on fusion research a long time ago
It still follows the laws of blackbody radiation - it's just that the emissivity of the body is part of the equation.
A classical blackbody has a emissivity of 1. This means not only that it absorbs radiation really well, it also means it's really good at radiating energy away.
Things that have low emissivity (all things transparent and all things reflective) are also really bad at radiating energy away. This is used for solar-thermal collectors today: you make them from an engineered material that is completely black at in the visible range, but highly reflective in the infra-red. That way, they absorb sunlight and get hot, but they don't lose heat energy because they can't radiate it away as heat radiation.
And yes, fusion plasma is extremely, extremely transparent. Not only is it extremely thin (ITER or Wendeltstein 7-X contain only 1-2g of hydrogen during operation), hydrogen is also extremely bad at absorbing gamma-rays (black body radiation at 1e8 K).
The heat loss is practically limited by the mass of hydrogen fusing in the machine. To have a continuous heat flux of 10^21 watts you would need to fuse ~4*10^5 kg of hydrogen every second. Which clearly these machines are not intended to do.
> Inertial confinement fusion like the NIF is not intended to run continuously, so the 2ns duration is irrelevant.
Indeed. I do think ICF has a future. The issue I described applies to machines that attempt to achieve sustained fusion. Pulsed fusion is ok.
> The heat loss is practically limited by the mass of hydrogen fusing in the machine.
Yes, but it goes the other way too. If the heat loss is to high you can't sustain fusion because you can't stay at the required temperature for long enough.
It would be a Lovecraftian nightmare of unmentionable proportions to actually operate, but you could imagine it breaking even.